I've been working this problem off and on for a month and decided that
since some of you use VIA Epia motherboards you might have seen
something like this before.
When I use our standard kickstart config or do a manual install and
specify that / and /boot is on a mirror'd volume and /home and the rest
are on a mirrored LVM volume I get this error:
An error occurred trying to format sysvg/home. This problem is serious,
and the install cannot continue.
Press <Enter> to reboot your system.
If I flip over to virtual console f3 I can see that it has successfully
formatted / and /boot and then tried to format /home
It shows the format command (all assembled, not in parameterized
function argument format as displayed on screen which is a pain to type) as:
/usr/sbin/mke3fs /dev/sysvg/home -i 4096 -j
This kickstart is off of PXE boot and with known good media (we have
installed a dozen other machines this way). I went through a manual
install after PXE booting the box and replicated our standard
partitioning scheme and had the exact same issue.
On several occasions while the processor calculates dependencies I have
gone to virtual console f2 and ran some commands to check the progress
and status of the install. The partitions all get created, the logical
volumes all get created, and everything looks good right up until it
produces the above error. After that no volume groups are active but
sysvg does exist. After the error I can go to f2 and re-activate the
volume groups and mkfs and access the filesystems. So I would guess that
something is happening to lvm at some point. I see no kernel oopsen,
nothing incriminating in dmesg. We are using software RAID. No RAID
card. md0, md1, and md2 exist after the error message is displayed but
md3 seems to have disappeared. lvm pvdisplay shows nothing but does
produce some messages such as "File descriptor 5 left open". Yesterday I
was able to manually create the md3 device and initialize it as a
physical volume and everything worked. Not sure why the installer is
having issues with it.
Scrolling up through the output on vc f4 I see that md3 did exist at one
point during the install.
I did see something about "cannot bd_claim" one of the disks but the
console has stopped scrolling for some reason. I have restarted the
install to see if I can get more info in that area.
This machine has a 1Ghz Nehemia CPU from VIA and a VIA mobo. It has 256M
of RAM and 2 Maxtor 80G ATA disks. One notable difference between this
machine and our usual kickstart is that all of our other stuff is SATA.
We did a search and replace on hda with sda etc everywhere to generate
this file to install our VIA machines with ATA disks. We have done this
with ATA disks before using this same kickstart modified the same way
with RHEL4 and had no issue.
RedHat has looked into this and used our kickstart config in a virtual
machine machine with 256M of RAM and 2 80G disks and the install worked
perfectly. So I'm guessing it has something to do with the hardware.
Anyone interested in all of the details can read the notes from our
RedHat support ticket which I have pasted here:
http://pastebin.ca/927325
Any ideas why I would be getting this strange error? I'm rather bummed
that even RedHat hasn't been able to figure this out in a month. If I
don't get this sorted in the next few days I am just going to set up the
machines without RAID. The plan is to run the two of them in HA mode
anyway. Was just hoping to have the extra protection.
--
Tracy R Reed Read my blog at http://ultraviolet.org
Key fingerprint = D4A8 4860 535C ABF8 BA97 25A6 F4F2 1829 9615 02AD
Non-GPG signed mail gets read only if I can find it among the spam.
--
[email protected]
http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list
